comparison of icd-10-am data quality between jurisdictions, as measured by picq 2002
DESCRIPTION
Comparison of ICD-10-AM Data Quality Between Jurisdictions, As Measured by PICQ 2002. Authors:Catherine Perry Sue Wood Kirsten McKenzie Andrea Groom Kerry Innes. What is PICQ 2002?. A software package For reviewing ICD-10-AM coded data Identifies coding variation in dataset - PowerPoint PPT PresentationTRANSCRIPT
Comparison of ICD-10-AM Data Quality Between Jurisdictions, As Measured by PICQ 2002Authors: Catherine Perry
Sue Wood Kirsten McKenzie Andrea Groom Kerry Innes
What is PICQ 2002?
A software package– For reviewing ICD-10-AM coded data– Identifies coding variation in dataset
Contains– Over 100 indicators: Fatal, Warning, Relative– Denominator / Numerator– Denominator count = records looked at– Numerator count = records with problem
Indicator Examples
• Records with rehabilitation care type but PDx code not admission for rehab
• Obstetric perineal laceration 1st/2nd degree without repair
• Hyperglycaemia code with diabetes code
• Diagnosis indicates death but separation mode is not death
• External Cause code required but not present with a trauma code
Aims of data analysis
• Determine areas where there are coding issues
• Highlight coding data quality issues of national significance (eg health priority areas)
• Inform:– Coder education– Amendments to ICD-10-AM– Development of PICQ
• Enhancements to current indicators• Indication as to what topics could be targeted for new
indicators
Data Analysed
AIHW Data ICD-10-AM Edition2000 Jan-Jun ICD-10-AM 1st Edition
2000 Jul-Dec ICD-10-AM 2nd Edition2001-02 Fin. Year ICD-10-AM 2nd Edition
2002-03 Fin. Year ICD-10-AM 3rd Edition
Methodology and Limitations
Used PICQ 2002 Software• Exclusion of some indicators due to:
– Data items not present in AIHW data– Problem since identified with individual PICQ
indicators
Analysis of data:• Basic results from PICQ data• Also some significance testing of results in SPSS
Not yet received all data
Results: Changes between Editions of ICD-10-AM1st edition to 2nd editionIn the first 6 months of
ICD-10-AM 2nd edition:
• 1 of 5 States improved their overall Fatal indicator rate
• 5 of 5 States improved their overall Warning indicator rate
2nd edition to 3rd editionIn the first 12 months of
ICD-10-AM 3rd edition:
• 3 of 5 States improved their overall Fatal indicator rate
• 4 of 5 States improved their overall Warning indicator rate
Results: Changes over time using ICD-10-AM 2nd Edition
Jul-Dec 2000 compared to 2001-02
• 4 of 5 States improved their overall Fatal indicator rate– State that did not improve went from having the
2nd highest rate in 2000, to a rate more than double any one else in 2001-02
• 4 of 5 States improve their overall Warning indicator rate– Less marked change than for Fatal indicators
Comparison of fatal indicator error rates between jurisdictions
0
0.5
1
1.5
2
2.5
3
A B C D E
Jurisdiction
Ind
ica
tor
Ra
te
2000 Jan-Jul 2000 Jul-Dec 2001/02 2002/03
Fatal Indicators with high rates
Indicator 1st Ed 2nd Ed 3rd Ed
Jan-Jun 2000 2001/02 2002/03
Secondary neoplasm site code without primary site code2.30 2.63 2.70
Alzheimer's disease code without dementia code8.56 11.00 9.68
Newborn affected by C/S w/out code indicating effect80.22 63.89 42.32
Type of spinal cord lesion w/out functional level of spinal cord lesion 54.49 27.21 26.41
Comparison of warning indicator error rates between jurisdictions
0
0.2
0.4
0.6
0.8
1
1.2
1.4
A B C D E
Jurisdiction
Ind
icato
r R
ate
2000 Jan-Jul 2000 Jul-Dec 2001/02 2002/03
Warning Indicators with high rates
Indicator 1st Ed 2nd Ed 3rd EdJan-Jun 2000 2001/02 2002/03
Cataract code as principal diagnosis before glaucoma code28.40 28.56 26.50
Insertion of pacemaker code without insertion of electrode code98.79 17.19 18.88
Diabetes mellitus, non insulin-dependant, age <30 years19.22 19.50 23.25
Respiratory distress of newborn unspecified21.20 18.67 15.43
Relationship between Fatal and Warning Indicators
0.0000
0.5000
1.0000
1.5000
2.0000
2.5000
A B C D E
Jurisdiction
Ind
ica
tor
Ra
tes
Fatal 2002/03 W arning 2002/03
What do the results say about coding quality in Australia?
Having a tool to measure the quality of coding has resulted in an overall improvement in these
areas nationally
Using PICQ at a Health Department level, and feeding the results back to hospitals, results in
improved data
That there is more work to be undertaken
Coding quality is not related to size of the jurisdiction
Indicator rates are just a starting point!
Great PICQ indicator rates great overall coding
Poor PICQ indicator rates poor overall coding
May highlight:– Coders not following coding standards &
conventions– Local coding practices– Classification issues– Documentation issues– Issues may be system related, rather than coder
related
Outcomes From PICQ AnalysisNCCH• Coder education (including in Coding
Matters)• Development of ICD-10-AM 4th Edition• Development of PICQ
Jurisdictions• Raised interest in benchmarking• Provided information to support education• Increased desire for information that can be
released to their Coding Committees
Where to from here?
Further analysis
Feed information back to Coding Committees for consideration at a local level
Release of PICQ 2004 (for use with 4th edition codes)– 97 New indicators for 4th edition– 223 Total indicators for 4th edition